Most color-television transmitters, including embodiments of the invention, send a composite signal comprising several components, among them a set of video components sufficient for the color picture. A related receiver decomposes the composite signal to extract the set of transmitted video components (or an equivalent set) which it then converts into a colored television display. Because the video components of an NTSC or related system are simple analog signals, they may be received with corrupting noise and distortions added thereto. The prior applications disclose transmitter means which quantize samples of analog video components in a manner which does not appreciably impair picture quality but facilitates noise removal in a receiver improved according to the invention.
Except where it is specifically stated otherwise, the set of transmitted components is assumed herein to comprise the luminance signal (designated Y) and the two chrominance signals (designated I and Q) of the NTSC system; likewise, a composite signal having the NTSC format is usually assumed hereinbelow. Persons having normal skill in the art will nevertheless understand that the principles of the invention apply equally to other systems, including the aforementioned SECAM, PAL and video recording systems of prior art.
The prior application, now Pat. No. 4,275,411, teaches very coarse quantization of samples of one or more of the video components after 3-dimensional dither has been added to prevent perceptible quality loss in the received television picture; this is called "dither coding (or dither quantizing) with 3-d dither." Coarse quantization is conveniently exemplified by four amplitude levels for the luminance and two levels for each chrominance component. When the I and Q components are severely quantized, very few resultant chromaticity values are transmitted. However, it will be clear to persons skilled in the art that 3-dimensional dithers vary from frame to frame the chromaticity sent for a picture element, and that temporal averaging of a sequence of frames produces the effect of intermediate chromaticities. Furthermore, mixtures of differently colored elements which appear within small picture areas also produce intermediate colors, due to subjective spatial averaging.
FIG. 1 illustrates dither coding of an analog signal X, which may be a video component, transmission of the dither-quantized signal over a real channel having conventional television bandwidth, and regeneration of the dither-coded signal. The interval of time shown in FIG. 1 corresponds to a small fraction of one horizontal sweep interval. Vertical grid lines represent instantaneous pulses of an ideal clock signal designated CLX and horizontal grid lines represent four quantum amplitude levels, labelled I, II, III and IV. (Y, I or Q may be substituted for the letter X.) FIG. 1A shows a portion of dither signal DX in relation to clock pulses CLX; the DX waveshape actually changes from line to line and from frame to frame of the scanning raster. Although DX is assumed to comprise NRZ (non-return-to-zero) pulses as shown by the solid line waveshape, narrower pulses, such as those of the dotted curve DX', would also be suitable, provided that the correct amplitude values prevail at all clock-pulse instants.
FIG. 1B represents an analog signal X by a dotted line and the sum X+DX by a solid line. For each clock-pulse instant, there occurs a quantized sample 6 (represented by a solid circle), at the intersection of the vertical grid line (clock pulse) with whichever of the quantum-level lines I, II, III or IV is closest to X+DX. FIG. 1C shows an ideal NRZ signal 3, obtained by storing each sample 6 during the ensuing clock interval. The solid line of FIG. 1D shows a distorted version 4 received when signal 3 is transmitted over a practicable channel. FIG. 1D also shows regenerated but delayed samples 6', obtained by resampling signal 4 midway between the vertical lines and requantizing to the I, II, III and IV levels. Resampling half a period late compensates for the average delay due to storing samples 6 of FIG. 1C for a full clock period. Persons skilled in the art will understand that, in a practical system, it may also be necessary to compensate for minor additional delays, not shown in FIG. 1D. The NRZ signal 3' is derived from samples 6' and is a delayed replica of 3.
In theory, all of the information in a stream of samples generated at the rate of 2W samples per sec. can be transmitted over an "ideal" channel having W Hz. bandwidth; a physically realizable channel, however, requires additional bandwidth. Therefore, the component video signals employed in accordance with the invention have bandwidths slightly less than those of the subchannels over which they are transmitted. Although the NTSC composite signal is designed to accommodate 4.2 MHZ. in the Y subchannel, 1.5 MHZ. in the I subchannel and 0.5 MHZ. in the Q subchannel, narrower signal bandwidths are tolerable and, indeed, are very common in the prior art.
In order to receive samples having the most accurate amplitude values, it is also necessary to equalize the response characteristics of the subchannels as taught, for example, in the book "Transmission Systems for Communications", published by Bell Telephone Laboratories (4th Edition, 1970, pp. 646-656 and 715-718 with references). In many cases, however, inasmuch as requantizing corrects for both channel distortion and noise, it is convenient to forego optimum equalization of each subchannel at a cost of some noise margin.
Best use of the invention requires the use of dithers (called 3-dimensional dithers) which characteristically vary from frame to frame of the television picture in a precise manner which greatly reduces the number of quantum levels needed. It is also important to avoid patterns of dither which produce frame-rate flicker on an interlaced scanning raster or which cause other noticeable picture impairments. Three-dimensional ordered dithers of the nasik species have these required properties and are otherwise convenient. They are therefore assumed in the descriptions herein of exemplary invention embodiments. Nasik dithers are described in my U.S. Pat. No. 3,739,082, entitled "Ordered Dither System".
Videodisk records, and recording and reproducing systems suitable therefore, are well known in the art and are described, for example, in the entire March 1978 issue of RCA Review, vol. 39, no. 1 (1978) and in Philips Technical Review, vol. 33, No. 7 (1973).
In order that descriptions of various invention embodiments may be facilitated by reference to generic quantizing means, FIG. 2 shows examples of such means suitable both for dither coding and to regenerate dither-coded signals. Assuming zero mean value for dither DX of FIGS. 1A and 1B, it is immaterial whether we dither-code by quantizing the sum X+DX, as shown in FIG. 1B, or the difference X-DX. For the invention embodiments disclosed hereinbelow, the quantity X.+-.DX is sampled as well as quantized, and sampling may occur before or after quantization, or even simultaneously therewith, in synchronism with a clock signal.
Means 415 of FIG. 2A produces binary samples (i.e. samples quantized with only two levels) and means 420 of FIG. 2C quantizes with multiple levels (four levels shown). Each of these means samples and quantizes substantially at the same time and also stores quantized samples so as to generate NRZ output waveshapes. Although NRZ outputs are not essential for the herein-disclosed invention embodiments, they are convenient.
Comparator 401 of FIG. 2A may be a commercial unit of the prior art, assumed to be an integrated-circuit device which includes "strobe" and "latch" functions. Analog signal X is furnished to comparator input 417, dither DX is provided to input 418, and clock signal CLX is furnished to the strobe input 419. While strobed by a clock pulse, unit 401 compares X with DX; if the sampled difference X-DX is negative (corresponding to the left-hand half of FIG. 2B), comparator output S(X-DX) goes "low" to quantum level I, and if the difference is positive (corresponding to the right-hand part of FIG. 2B), S(X-DX) goes "high" to quantum level II. Since the circuit also latches, the output remains at either of these levels between clock pulses, resulting in an NRZ binary signal.
Assuming that dither DX has already been added to analog signal X, FIG. 2C shows 4-level quantizer 420 which samples and quantizes X+DX when clock signal CLX is furnished to lead 411, the strobe input. Common lead 407 furnishes X+DX to one input of each of comparators 403, 404, and 405, which are similar to 401 of FIG. 2A. The second input 408 of comparator 403 is held constant at a reference voltage +e, the second input 409 of comparator 404 is held at zero, and the second input 410 of comparator 405 is held at -e. Each comparator output goes either high or low, depending on the value of X+DX, and the three outputs are added in summing amplifier 406, producing a 4-level quantization characteristic F(X+DX), shown as a dotted-line graph in FIG. 2B.
If X+DX is more negative than -e when a clock pulse occurs, all three comparator outputs go to the low state and remain there until the next clock pulse; F(X+DX) then corresponds to quantum level I. When X+DX is between -e and zero, only output 414 from comparator 405 rises while the other two outputs remain low; F(X+DX) then rests at quantum level II. In similar fashion, level III results when X+DX is between zero and +e, and level IV when the input is more positive than +e. Because the comparators latch F(X+DX) is put out as an NRZ signal.
An already-quantized 4-level signal, adjusted so that levels, I, II, III and IV respectively correspond to approximately -3/2e, -1/2e, +1/2e, and +3/2e, may be resampled and requantized by applying it to input lead 407 of quantizer 420; this the regeneration process which removes noise. In similar fashion, a binary signal having equally positive and negative quantum levels may be regenerated with 2-level quantizer 415, lead 418 being held at zero potential.
It will be apparent to persons skilled in the art that -e, zero, and +e are quantizer decision levels and that, more generally, an N-level quantizer can be constructed in the manner of 420 of FIG. 2C, using N-1 comparators and N-1 decision levels. Furthermore, the -e, zero and +e voltages shown in FIGS. 2A, 2B, and 2C may all be augmented by the same additional voltage, and the high and low output levels of binary quantizer 415 need not be the same as any output levels of multilevel quantizer 420.
It will also be apparent to persons skilled in the art how quantizer 420 is equivalent to, and may be replaced by, an analog-to-digital (a/d) converter followed by a digital-to-analog (d/a) converter, such as 508 and 510 of FIG. 13. It can also be implemented with the aid of level-detector devices of commerce and various other staircase-quantizer means of prior art.